FastGCN icon indicating copy to clipboard operation
FastGCN copied to clipboard

Why not using the GPUs to train ?

Open MrLinNing opened this issue 4 years ago • 1 comments

Hello, you did a wonderful job, and why not using GPUs to train ? Does the training procedure of GPU is slower than that of CPU?

MrLinNing avatar Oct 28 '19 06:10 MrLinNing

Early experiments were done using CPU only, so we just reported those results. We tried 1 GPU on PubMed data before, it is comparable to but slightly slower than 1 CPU. But I think it is not concluded. The performance depends on the experimental environment.

matenure avatar Oct 28 '19 21:10 matenure